summaryrefslogtreecommitdiffstats
path: root/libFDK/src/arm/qmf_arm.cpp
blob: 39795d7bc4528ac58814916817709e4a0ecd0b4f (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
/* -----------------------------------------------------------------------------------------------------------
Software License for The Fraunhofer FDK AAC Codec Library for Android

� Copyright  1995 - 2012 Fraunhofer-Gesellschaft zur F�rderung der angewandten Forschung e.V.
  All rights reserved.

 1.    INTRODUCTION
The Fraunhofer FDK AAC Codec Library for Android ("FDK AAC Codec") is software that implements
the MPEG Advanced Audio Coding ("AAC") encoding and decoding scheme for digital audio.
This FDK AAC Codec software is intended to be used on a wide variety of Android devices.

AAC's HE-AAC and HE-AAC v2 versions are regarded as today's most efficient general perceptual
audio codecs. AAC-ELD is considered the best-performing full-bandwidth communications codec by
independent studies and is widely deployed. AAC has been standardized by ISO and IEC as part
of the MPEG specifications.

Patent licenses for necessary patent claims for the FDK AAC Codec (including those of Fraunhofer)
may be obtained through Via Licensing (www.vialicensing.com) or through the respective patent owners
individually for the purpose of encoding or decoding bit streams in products that are compliant with
the ISO/IEC MPEG audio standards. Please note that most manufacturers of Android devices already license
these patent claims through Via Licensing or directly from the patent owners, and therefore FDK AAC Codec
software may already be covered under those patent licenses when it is used for those licensed purposes only.

Commercially-licensed AAC software libraries, including floating-point versions with enhanced sound quality,
are also available from Fraunhofer. Users are encouraged to check the Fraunhofer website for additional
applications information and documentation.

2.    COPYRIGHT LICENSE

Redistribution and use in source and binary forms, with or without modification, are permitted without
payment of copyright license fees provided that you satisfy the following conditions:

You must retain the complete text of this software license in redistributions of the FDK AAC Codec or
your modifications thereto in source code form.

You must retain the complete text of this software license in the documentation and/or other materials
provided with redistributions of the FDK AAC Codec or your modifications thereto in binary form.
You must make available free of charge copies of the complete source code of the FDK AAC Codec and your
modifications thereto to recipients of copies in binary form.

The name of Fraunhofer may not be used to endorse or promote products derived from this library without
prior written permission.

You may not charge copyright license fees for anyone to use, copy or distribute the FDK AAC Codec
software or your modifications thereto.

Your modified versions of the FDK AAC Codec must carry prominent notices stating that you changed the software
and the date of any change. For modified versions of the FDK AAC Codec, the term
"Fraunhofer FDK AAC Codec Library for Android" must be replaced by the term
"Third-Party Modified Version of the Fraunhofer FDK AAC Codec Library for Android."

3.    NO PATENT LICENSE

NO EXPRESS OR IMPLIED LICENSES TO ANY PATENT CLAIMS, including without limitation the patents of Fraunhofer,
ARE GRANTED BY THIS SOFTWARE LICENSE. Fraunhofer provides no warranty of patent non-infringement with
respect to this software.

You may use this FDK AAC Codec software or modifications thereto only for purposes that are authorized
by appropriate patent licenses.

4.    DISCLAIMER

This FDK AAC Codec software is provided by Fraunhofer on behalf of the copyright holders and contributors
"AS IS" and WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES, including but not limited to the implied warranties
of merchantability and fitness for a particular purpose. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
CONTRIBUTORS BE LIABLE for any direct, indirect, incidental, special, exemplary, or consequential damages,
including but not limited to procurement of substitute goods or services; loss of use, data, or profits,
or business interruption, however caused and on any theory of liability, whether in contract, strict
liability, or tort (including negligence), arising in any way out of the use of this software, even if
advised of the possibility of such damage.

5.    CONTACT INFORMATION

Fraunhofer Institute for Integrated Circuits IIS
Attention: Audio and Multimedia Departments - FDK AAC LL
Am Wolfsmantel 33
91058 Erlangen, Germany

www.iis.fraunhofer.de/amm
amm-info@iis.fraunhofer.de
----------------------------------------------------------------------------------------------------------- */

#if (QMF_NO_POLY==5)

#define FUNCTION_qmfForwardModulationLP_odd

#ifdef FUNCTION_qmfForwardModulationLP_odd
static void
qmfForwardModulationLP_odd( HANDLE_QMF_FILTER_BANK anaQmf, /*!< Handle of Qmf Analysis Bank  */
                            const FIXP_QMF *timeIn,        /*!< Time Signal */
                            FIXP_QMF *rSubband )           /*!< Real Output */
{
  int i;
  int L = anaQmf->no_channels;
  int M = L>>1;
  int shift = (anaQmf->no_channels>>6) + 1;
  int rSubband_e = 0;

  FIXP_QMF *rSubbandPtr0 = &rSubband[M+0];                /* runs with increment */
  FIXP_QMF *rSubbandPtr1 = &rSubband[M-1];                /* runs with decrement */
  FIXP_QMF *timeIn0 = (FIXP_DBL *) &timeIn[0];            /* runs with increment */
  FIXP_QMF *timeIn1 = (FIXP_DBL *) &timeIn[L];            /* runs with increment */
  FIXP_QMF *timeIn2 = (FIXP_DBL *) &timeIn[L-1];          /* runs with decrement */
  FIXP_QMF *timeIn3 = (FIXP_DBL *) &timeIn[2*L-1];        /* runs with decrement */

  for (i = 0; i < M; i++)
  {
    *rSubbandPtr0++ = (*timeIn2-- >> 1) - (*timeIn0++ >> shift);
    *rSubbandPtr1-- = (*timeIn1++ >> 1) + (*timeIn3-- >> shift);
  }

  dct_IV(rSubband,L, &rSubband_e);
}
#endif /* FUNCTION_qmfForwardModulationLP_odd */


/* NEON optimized QMF currently builts only with RVCT toolchain */

#if defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_5TE__)

#if (SAMPLE_BITS == 16)
#define FUNCTION_qmfAnaPrototypeFirSlot
#endif

#ifdef FUNCTION_qmfAnaPrototypeFirSlot

#if defined(__GNUC__)	/* cppp replaced: elif */

inline INT SMULBB (const SHORT a, const LONG b)
{
  INT result ;
  __asm__ ("smulbb %0, %1, %2"
     : "=r" (result)
     : "r" (a), "r" (b)) ;
  return result ;
}
inline INT SMULBT (const SHORT a, const LONG b)
{
  INT result ;
  __asm__ ("smulbt %0, %1, %2"
     : "=r" (result)
     : "r" (a), "r" (b)) ;
  return result ;
}

inline INT SMLABB(const LONG accu, const SHORT a, const LONG b)
{
  INT result ;
  __asm__ ("smlabb %0, %1, %2,%3"
     : "=r" (result)
     : "r" (a), "r" (b), "r" (accu)) ;
  return result;
}
inline INT SMLABT(const LONG accu, const SHORT a, const LONG b)
{
  INT result ;
  __asm__ ("smlabt %0, %1, %2,%3"
     : "=r" (result)
     : "r" (a), "r" (b), "r" (accu)) ;
  return result;
}
#endif /* compiler selection  */


void qmfAnaPrototypeFirSlot( FIXP_QMF *analysisBuffer,
                             int       no_channels,             /*!< Number channels of analysis filter */
                             const FIXP_PFT *p_filter,
                             int       p_stride,                /*!< Stide of analysis filter    */
                             FIXP_QAS *RESTRICT pFilterStates
                            )
{
  LONG *p_flt = (LONG *) p_filter;
  LONG flt;
  FIXP_QMF *RESTRICT pData_0 = analysisBuffer + 2*no_channels - 1;
  FIXP_QMF *RESTRICT pData_1 = analysisBuffer;

  FIXP_QAS *RESTRICT sta_0 = (FIXP_QAS *)pFilterStates;
  FIXP_QAS *RESTRICT sta_1 = (FIXP_QAS *)pFilterStates + (2*QMF_NO_POLY*no_channels) - 1;

  FIXP_DBL accu0, accu1;
  FIXP_QAS sta0, sta1;

  int staStep1 =  no_channels<<1;
  int staStep2 = (no_channels<<3) - 1; /* Rewind one less */

  if (p_stride == 1)
  {
    /* FIR filter 0 */
    flt = *p_flt++;
    sta1 = *sta_1;  sta_1 -= staStep1;
    accu1 = SMULBB(        sta1, flt);
    sta1 = *sta_1;  sta_1 -= staStep1;
    accu1 = SMLABT( accu1, sta1, flt);

    flt = *p_flt++;
    sta1 = *sta_1;  sta_1 -= staStep1;
    accu1 = SMLABB( accu1, sta1, flt);
    sta1 = *sta_1;  sta_1 -= staStep1;
    accu1 = SMLABT( accu1, sta1, flt);

    flt = *p_flt++;
    sta1 = *sta_1;  sta_1 += staStep2;
    accu1 = SMLABB( accu1, sta1, flt);
    *pData_1++ = FX_DBL2FX_QMF(accu1<<1);

    /* FIR filters 1..63 127..65 or 1..31 63..33 */
    no_channels >>= 1;
    for (; --no_channels; )
    {
      sta0 = *sta_0; sta_0 += staStep1;  /* 1,3,5, ... 29/61 */
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMULBT(        sta0, flt);
      accu1 = SMULBT(        sta1, flt);

      flt = *p_flt++;
      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMLABB( accu0, sta0, flt);
      accu1 = SMLABB( accu1, sta1, flt);

      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMLABT( accu0, sta0, flt);
      accu1 = SMLABT( accu1, sta1, flt);

      flt = *p_flt++;
      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMLABB( accu0, sta0, flt);
      accu1 = SMLABB( accu1, sta1, flt);

      sta0 = *sta_0; sta_0 -= staStep2;
      sta1 = *sta_1; sta_1 += staStep2;
      accu0 = SMLABT( accu0, sta0, flt);
      accu1 = SMLABT( accu1, sta1, flt);

      *pData_0-- = FX_DBL2FX_QMF(accu0<<1);
      *pData_1++ = FX_DBL2FX_QMF(accu1<<1);

      /* Same sequence as above, but mix B=bottom with T=Top */

      flt = *p_flt++;
      sta0 = *sta_0; sta_0 += staStep1;  /* 2,4,6, ... 30/62 */
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMULBB(        sta0, flt);
      accu1 = SMULBB(        sta1, flt);

      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMLABT( accu0, sta0, flt);
      accu1 = SMLABT( accu1, sta1, flt);

      flt = *p_flt++;
      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMLABB( accu0, sta0, flt);
      accu1 = SMLABB( accu1, sta1, flt);

      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMLABT( accu0, sta0, flt);
      accu1 = SMLABT( accu1, sta1, flt);

      flt = *p_flt++;
      sta0 = *sta_0; sta_0 -= staStep2;
      sta1 = *sta_1; sta_1 += staStep2;
      accu0 = SMLABB( accu0, sta0, flt);
      accu1 = SMLABB( accu1, sta1, flt);

      *pData_0-- = FX_DBL2FX_QMF(accu0<<1);
      *pData_1++ = FX_DBL2FX_QMF(accu1<<1);
    }

    /* FIR filter 31/63 and 33/65 */
    sta0 = *sta_0; sta_0 += staStep1;
    sta1 = *sta_1; sta_1 -= staStep1;
    accu0 = SMULBT(        sta0, flt);
    accu1 = SMULBT(        sta1, flt);

    flt = *p_flt++;
    sta0 = *sta_0; sta_0 += staStep1;
    sta1 = *sta_1; sta_1 -= staStep1;
    accu0 = SMLABB( accu0, sta0, flt);
    accu1 = SMLABB( accu1, sta1, flt);

    sta0 = *sta_0; sta_0 += staStep1;
    sta1 = *sta_1; sta_1 -= staStep1;
    accu0 = SMLABT( accu0, sta0, flt);
    accu1 = SMLABT( accu1, sta1, flt);

    flt = *p_flt++;
    sta0 = *sta_0; sta_0 += staStep1;
    sta1 = *sta_1; sta_1 -= staStep1;
    accu0 = SMLABB( accu0, sta0, flt);
    accu1 = SMLABB( accu1, sta1, flt);

    sta0 = *sta_0; sta_0 -= staStep2;
    sta1 = *sta_1; sta_1 += staStep2;
    accu0 = SMLABT( accu0, sta0, flt);
    accu1 = SMLABT( accu1, sta1, flt);

    *pData_0-- = FX_DBL2FX_QMF(accu0<<1);
    *pData_1++ = FX_DBL2FX_QMF(accu1<<1);

    /* FIR filter 32/64 */
    flt = *p_flt++;
    sta0 = *sta_0; sta_0 += staStep1;
    sta1 = *sta_1; sta_1 -= staStep1;
    accu0 = SMULBB(        sta0, flt);
    accu1 = SMULBB(        sta1, flt);

    sta0 = *sta_0; sta_0 += staStep1;
    sta1 = *sta_1; sta_1 -= staStep1;
    accu0 = SMLABT( accu0, sta0, flt);
    accu1 = SMLABT( accu1, sta1, flt);

    flt = *p_flt++;
    sta0 = *sta_0; sta_0 += staStep1;
    sta1 = *sta_1; sta_1 -= staStep1;
    accu0 = SMLABB( accu0, sta0, flt);
    accu1 = SMLABB( accu1, sta1, flt);

    sta0 = *sta_0; sta_0 += staStep1;
    sta1 = *sta_1; sta_1 -= staStep1;
    accu0 = SMLABT( accu0, sta0, flt);
    accu1 = SMLABT( accu1, sta1, flt);

    flt = *p_flt;
    sta0 = *sta_0;
    sta1 = *sta_1;
    accu0 = SMLABB( accu0, sta0, flt);
    accu1 = SMLABB( accu1, sta1, flt);

    *pData_0-- = FX_DBL2FX_QMF(accu0<<1);
    *pData_1++ = FX_DBL2FX_QMF(accu1<<1);
  }
  else
  {
    int pfltStep = QMF_NO_POLY * (p_stride-1);

    flt = p_flt[0];
    sta1 = *sta_1;  sta_1 -= staStep1;
    accu1 = SMULBB(        sta1, flt);
    sta1 = *sta_1;  sta_1 -= staStep1;
    accu1 = SMLABT( accu1, sta1, flt);

    flt = p_flt[1];
    sta1 = *sta_1;  sta_1 -= staStep1;
    accu1 = SMLABB( accu1, sta1, flt);
    sta1 = *sta_1;  sta_1 -= staStep1;
    accu1 = SMLABT( accu1, sta1, flt);

    flt = p_flt[2]; p_flt += pfltStep;
    sta1 = *sta_1;  sta_1 += staStep2;
    accu1 = SMLABB( accu1, sta1, flt);
    *pData_1++ = FX_DBL2FX_QMF(accu1<<1);

    /* FIR filters 1..63 127..65 or 1..31 63..33 */
    for (; --no_channels; )
    {
      flt = p_flt[0];
      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMULBB(        sta0, flt);
      accu1 = SMULBB(        sta1, flt);

      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMLABT( accu0, sta0, flt);
      accu1 = SMLABT( accu1, sta1, flt);

      flt = p_flt[1];
      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMLABB( accu0, sta0, flt);
      accu1 = SMLABB( accu1, sta1, flt);

      sta0 = *sta_0; sta_0 += staStep1;
      sta1 = *sta_1; sta_1 -= staStep1;
      accu0 = SMLABT( accu0, sta0, flt);
      accu1 = SMLABT( accu1, sta1, flt);

      flt = p_flt[2]; p_flt += pfltStep;
      sta0 = *sta_0; sta_0 -= staStep2;
      sta1 = *sta_1; sta_1 += staStep2;
      accu0 = SMLABB( accu0, sta0, flt);
      accu1 = SMLABB( accu1, sta1, flt);

      *pData_0-- = FX_DBL2FX_QMF(accu0<<1);
      *pData_1++ = FX_DBL2FX_QMF(accu1<<1);
    }

    /* FIR filter 32/64 */
    flt = p_flt[0];
    sta0 = *sta_0; sta_0 += staStep1;
    accu0 = SMULBB(        sta0, flt);
    sta0 = *sta_0; sta_0 += staStep1;
    accu0 = SMLABT( accu0, sta0, flt);

    flt = p_flt[1];
    sta0 = *sta_0; sta_0 += staStep1;
    accu0 = SMLABB( accu0, sta0, flt);
    sta0 = *sta_0; sta_0 += staStep1;
    accu0 = SMLABT( accu0, sta0, flt);

    flt = p_flt[2];
    sta0 = *sta_0;
    accu0 = SMLABB( accu0, sta0, flt);
    *pData_0-- = FX_DBL2FX_QMF(accu0<<1);
  }
}
#endif /* FUNCTION_qmfAnaPrototypeFirSlot */
#endif /* #if defined(__CC_ARM) && defined(__ARM_ARCH_6__) */

#if ( defined(__ARM_ARCH_5TE__) && (SAMPLE_BITS == 16) ) && !defined(QMF_TABLE_FULL)

#define FUNCTION_qmfSynPrototypeFirSlot

#if defined(FUNCTION_qmfSynPrototypeFirSlot)

#if defined(__GNUC__)	/* cppp replaced: elif */

inline INT SMULWB (const LONG a, const LONG b)
{
  INT result ;
  __asm__ ("smulwb %0, %1, %2"
    : "=r" (result)
    : "r" (a), "r" (b)) ;

  return result ;
}
inline INT SMULWT (const LONG a, const LONG b)
{
  INT result ;
  __asm__ ("smulwt %0, %1, %2"
    : "=r" (result)
    : "r" (a), "r" (b)) ;

  return result ;
}

inline INT SMLAWB(const LONG accu, const LONG a, const LONG b)
{
  INT result;
  asm("smlawb %0, %1, %2, %3 "
        : "=r" (result)
        : "r" (a), "r" (b), "r" (accu) );
  return result ;
}

inline INT SMLAWT(const LONG accu, const LONG a, const LONG b)
{
  INT result;
  asm("smlawt %0, %1, %2, %3 "
        : "=r" (result)
        : "r" (a), "r" (b), "r" (accu) );
  return result ;
}

#endif /* ARM compiler selector */


static void qmfSynPrototypeFirSlot1_filter(FIXP_QMF *RESTRICT realSlot, 
                                           FIXP_QMF *RESTRICT imagSlot, 
                                           const FIXP_DBL *RESTRICT p_flt, 
                                           FIXP_QSS *RESTRICT sta,
                                           FIXP_DBL *pMyTimeOut, 
                                           int no_channels)
{
  /* This code was the base for the above listed assembler sequence */
  /* It can be used for debugging purpose or further optimizations  */
  const FIXP_DBL *RESTRICT p_fltm = p_flt + 155;

  do
  {
     FIXP_DBL result;
     FIXP_DBL A, B, real, imag, sta0;

     real = *--realSlot;
     imag = *--imagSlot;
     B = p_flt[4];                        /* Bottom=[8] Top=[9]     */
     A = p_fltm[3];                       /* Bottom=[316] Top=[317] */
     sta0 = sta[0];                       /* save state[0]          */
     *sta++ = SMLAWT( sta[1], imag, B );  /* index=9...........319  */
     *sta++ = SMLAWB( sta[1], real, A );  /* index=316...........6  */
     *sta++ = SMLAWB( sta[1], imag, B );  /* index=8,18,    ...318  */
     B = p_flt[3];                        /* Bottom=[6] Top=[7]     */
     *sta++ = SMLAWT( sta[1], real, A );  /* index=317...........7  */
     A = p_fltm[4];                       /* Bottom=[318] Top=[319] */
     *sta++ = SMLAWT( sta[1], imag, B );  /* index=7...........317  */
     *sta++ = SMLAWB( sta[1], real, A );  /* index=318...........8  */
     *sta++ = SMLAWB( sta[1], imag, B );  /* index=6...........316  */
     B = p_flt[2];                        /* Bottom=[X] Top=[5]     */
     *sta++ = SMLAWT( sta[1], real, A );  /* index=9...........319  */
     A = p_fltm[2];                       /* Bottom=[X] Top=[315]   */
     *sta++ =         SMULWT( imag, B );  /* index=5,15, ...   315  */
     result = SMLAWT( sta0,   real, A );  /* index=315...........5  */

     *pMyTimeOut++ = result;

     real = *--realSlot;
     imag = *--imagSlot;
     A = p_fltm[0];                       /* Bottom=[310] Top=[311] */
     B = p_flt[7];                        /* Bottom=[14]  Top=[15]  */
     result = SMLAWB( sta[0], real, A );  /* index=310...........0  */
     *sta++ = SMLAWB( sta[1], imag, B );  /* index=14..........324  */
     *pMyTimeOut++ = result;
     B = p_flt[6];                        /* Bottom=[12]  Top=[13]  */
     *sta++ = SMLAWT( sta[1], real, A );  /* index=311...........1  */
     A = p_fltm[1];                       /* Bottom=[312] Top=[313] */
     *sta++ = SMLAWT( sta[1], imag, B );  /* index=13..........323  */
     *sta++ = SMLAWB( sta[1], real, A );  /* index=312...........2  */
     *sta++ = SMLAWB( sta[1], imag, B );  /* index=12..........322  */
     *sta++ = SMLAWT( sta[1], real, A );  /* index=313...........3  */
     A = p_fltm[2];                       /* Bottom=[314] Top=[315] */
     B = p_flt[5];                        /* Bottom=[10]  Top=[11]  */
     *sta++ = SMLAWT( sta[1], imag, B );  /* index=11..........321  */
     *sta++ = SMLAWB( sta[1], real, A );  /* index=314...........4  */
     *sta++ =         SMULWB( imag, B );  /* index=10..........320  */


     p_flt    += 5;
     p_fltm   -= 5;
  } 
  while ((--no_channels) != 0);

}



INT qmfSynPrototypeFirSlot2(
                             HANDLE_QMF_FILTER_BANK qmf,
                             FIXP_QMF *RESTRICT realSlot,            /*!< Input: Pointer to real Slot */
                             FIXP_QMF *RESTRICT imagSlot,            /*!< Input: Pointer to imag Slot */
                             INT_PCM  *RESTRICT timeOut,             /*!< Time domain data */
                             INT       stride                        /*!< Time output buffer stride factor*/
                            )
{
  FIXP_QSS *RESTRICT sta = (FIXP_QSS*)qmf->FilterStates;
  int no_channels = qmf->no_channels;
  int scale = ((DFRACT_BITS-SAMPLE_BITS)-1-qmf->outScalefactor);

  /* We map an arry of 16-bit values upon an array of 2*16-bit values to read 2 values in one shot */
  const FIXP_DBL *RESTRICT p_flt  = (FIXP_DBL *) qmf->p_filter;           /* low=[0],   high=[1]   */
  const FIXP_DBL *RESTRICT p_fltm = (FIXP_DBL *) qmf->p_filter + 155;     /* low=[310], high=[311] */

  FDK_ASSERT(SAMPLE_BITS-1-qmf->outScalefactor >= 0); //   (DFRACT_BITS-SAMPLE_BITS)-1-qmf->outScalefactor >= 0);
  FDK_ASSERT(qmf->p_stride==2 && qmf->no_channels == 32);

  FDK_ASSERT((no_channels&3) == 0);  /* should be a multiple of 4 */

  realSlot += no_channels-1;    // ~~"~~
  imagSlot += no_channels-1;    // no_channels-1 .. 0

  FIXP_DBL MyTimeOut[32];
  FIXP_DBL *pMyTimeOut = &MyTimeOut[0];

  for (no_channels = no_channels; no_channels--;)
  {
     FIXP_DBL result;
     FIXP_DBL A, B, real, imag;

     real = *realSlot--;
     imag = *imagSlot--;
     A = p_fltm[0];                       /* Bottom=[310] Top=[311] */
     B = p_flt[7];                        /* Bottom=[14]  Top=[15]  */
     result = SMLAWB( sta[0], real, A );  /* index=310...........0  */
     *sta++ = SMLAWB( sta[1], imag, B );  /* index=14..........324  */
     B = p_flt[6];                        /* Bottom=[12]  Top=[13]  */
     *sta++ = SMLAWT( sta[1], real, A );  /* index=311...........1  */
     A = p_fltm[1];                       /* Bottom=[312] Top=[313] */
     *sta++ = SMLAWT( sta[1], imag, B );  /* index=13..........323  */
     *sta++ = SMLAWB( sta[1], real, A );  /* index=312...........2  */
     *sta++ = SMLAWB( sta[1], imag, B );  /* index=12..........322  */
     *sta++ = SMLAWT( sta[1], real, A );  /* index=313...........3  */
     A = p_fltm[2];                       /* Bottom=[314] Top=[315] */
     B = p_flt[5];                        /* Bottom=[10]  Top=[11]  */
     *sta++ = SMLAWT( sta[1], imag, B );  /* index=11..........321  */
     *sta++ = SMLAWB( sta[1], real, A );  /* index=314...........4  */
     *sta++ =         SMULWB( imag, B );  /* index=10..........320  */

     *pMyTimeOut++ = result;

     p_fltm   -= 5;
     p_flt    += 5;
  }

  pMyTimeOut = &MyTimeOut[0];
#if (SAMPLE_BITS == 16)      
  const FIXP_DBL max_pos = (FIXP_DBL) 0x00007FFF << scale;
  const FIXP_DBL max_neg = (FIXP_DBL) 0xFFFF8001 << scale;
#else
  scale = -scale;
  const FIXP_DBL max_pos = (FIXP_DBL) 0x7FFFFFFF >> scale;
  const FIXP_DBL max_neg = (FIXP_DBL) 0x80000001 >> scale;  
#endif
  const FIXP_DBL add_neg = (1 << scale) - 1;

  no_channels = qmf->no_channels;

  timeOut += no_channels*stride;

  FDK_ASSERT(scale >= 0);

  if (qmf->outGain != 0x80000000)
  {
    FIXP_DBL gain = qmf->outGain;
    for (no_channels>>=2; no_channels--;)
    {
      FIXP_DBL result1, result2;

      result1 = *pMyTimeOut++;
      result2 = *pMyTimeOut++;

      result1 = fMult(result1,gain);
      timeOut -= stride;
      if (result1 < 0)        result1 += add_neg;
      if (result1 < max_neg)  result1 = max_neg;
      if (result1 > max_pos)  result1 = max_pos;
#if (SAMPLE_BITS == 16)
      timeOut[0] = result1 >> scale;
#else
      timeOut[0] = result1 << scale;
#endif

      result2 = fMult(result2,gain);
      timeOut -= stride;
      if (result2 < 0)        result2 += add_neg;
      if (result2 < max_neg)  result2 = max_neg;
      if (result2 > max_pos)  result2 = max_pos;
#if (SAMPLE_BITS == 16)
      timeOut[0] = result2 >> scale;
#else
      timeOut[0] = result2 << scale;
#endif

      result1 = *pMyTimeOut++;
      result2 = *pMyTimeOut++;

      result1 = fMult(result1,gain);
      timeOut -= stride;
      if (result1 < 0)        result1 += add_neg;
      if (result1 < max_neg)  result1 = max_neg;
      if (result1 > max_pos)  result1 = max_pos;
#if (SAMPLE_BITS == 16)
      timeOut[0] = result1 >> scale;
#else
      timeOut[0] = result1 << scale;
#endif

      result2 = fMult(result2,gain);
      timeOut -= stride;
      if (result2 < 0)        result2 += add_neg;
      if (result2 < max_neg)  result2 = max_neg;
      if (result2 > max_pos)  result2 = max_pos;
#if (SAMPLE_BITS == 16)
      timeOut[0] = result2 >> scale;
#else
      timeOut[0] = result2 << scale;
#endif
    }
  }
  else
  {
    for (no_channels>>=2; no_channels--;)
    {
      FIXP_DBL result1, result2;
      result1 = *pMyTimeOut++;
      result2 = *pMyTimeOut++;
      timeOut -= stride;
      if (result1 < 0)        result1 += add_neg;
      if (result1 < max_neg)  result1 = max_neg;
      if (result1 > max_pos)  result1 = max_pos;
#if (SAMPLE_BITS == 16)
      timeOut[0] = result1 >> scale;
#else
      timeOut[0] = result1 << scale;
#endif
      
      timeOut -= stride;
      if (result2 < 0)        result2 += add_neg;
      if (result2 < max_neg)  result2 = max_neg;
      if (result2 > max_pos)  result2 = max_pos;
#if (SAMPLE_BITS == 16)
      timeOut[0] = result2 >> scale;
#else
      timeOut[0] = result2 << scale;
#endif
      
      result1 = *pMyTimeOut++;
      result2 = *pMyTimeOut++;
      timeOut -= stride;
      if (result1 < 0)        result1 += add_neg;
      if (result1 < max_neg)  result1 = max_neg;
      if (result1 > max_pos)  result1 = max_pos;
#if (SAMPLE_BITS == 16)
      timeOut[0] = result1 >> scale;
#else
      timeOut[0] = result1 << scale;
#endif
      
      timeOut -= stride;
      if (result2 < 0)        result2 += add_neg;
      if (result2 < max_neg)  result2 = max_neg;
      if (result2 > max_pos)  result2 = max_pos;
#if (SAMPLE_BITS == 16)
      timeOut[0] = result2 >> scale;
#else
      timeOut[0] = result2 << scale;
#endif
    }
  }
  return 0;
}

static
void qmfSynPrototypeFirSlot_fallback( HANDLE_QMF_FILTER_BANK qmf,
                             FIXP_DBL *realSlot,      /*!< Input: Pointer to real Slot */
                             FIXP_DBL *imagSlot,      /*!< Input: Pointer to imag Slot */
                             INT_PCM  *timeOut,             /*!< Time domain data */
                             const int       stride
                            );

/*!
  \brief Perform Synthesis Prototype Filtering on a single slot of input data.

  The filter takes 2 * #MAX_SYNTHESIS_CHANNELS of input data and
  generates #MAX_SYNTHESIS_CHANNELS time domain output samples.
*/

static
void qmfSynPrototypeFirSlot( HANDLE_QMF_FILTER_BANK qmf,
                             FIXP_DBL *realSlot,      /*!< Input: Pointer to real Slot */
                             FIXP_DBL *imagSlot,      /*!< Input: Pointer to imag Slot */
                             INT_PCM  *timeOut,             /*!< Time domain data */
                             const int       stride
                            )
{
    INT err = -1;

    switch (qmf->p_stride) {
    case 2:
      err = qmfSynPrototypeFirSlot2(qmf, realSlot, imagSlot, timeOut, stride);
      break;
    default:
      err = -1;
    }

    /* fallback if configuration not available or failed */
    if(err!=0) {
        qmfSynPrototypeFirSlot_fallback(qmf, realSlot, imagSlot, timeOut, stride);
    }
}
#endif /* FUNCTION_qmfSynPrototypeFirSlot */

#endif  /*  ( defined(__CC_ARM) && defined(__ARM_ARCH_5TE__) && (SAMPLE_BITS == 16) ) && !defined(QMF_TABLE_FULL) */



/* #####################################################################################*/



#endif  /* (QMF_NO_POLY==5) */